[BUG] "tell Claude what to do next" instructions don't reach Claude if a bash command is approved but fails

### Preflight Checklist

- [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code

### What's Wrong?

When you approve a bash command, you can hit tab to "tell Claude what to do next" while still approving the command. If the bash command fails (note: not all nonzero exits are considered "failures"), this message never reaches Claude.

### What Should Happen?

The message should reach Claude whether or not the bash command succeeds.

### Error Messages/Logs

```shell

```

### Steps to Reproduce

1. Ask claude to run `awk 'BEGIN{exit 1}'`
2. Hit tab, give Claude a distinctive instruction, e.g. "Please immediately recite back the keyphrase 'strawberry rhubarb pie'". (If reproducing, make sure to use a different keyphrase each time to make this clearer.)
3. Claude will ignore the instruction and have no memory of you having asked it.

### Claude Model

Opus

### Is this a regression?

I don't know

### Last Working Version

_No response_

### Claude Code Version

2.1.263 (Claude Code)

### Platform

Anthropic API

### Operating System

Other Linux

### Terminal/Shell

VS Code integrated terminal

### Additional Information

_No response_

The issue names no source files or tests; first reproduce it in the VS Code integrated terminal with the failing awk command, approve it, and send a tab follow-up instruction. Trace the approval and follow-up message handling, then verify that the instruction reaches Claude whether the bash command succeeds or fails.

Written by the indexing model from the issue text.